Validation
Home Introduction Methodology Testing Output Plots Validation Future Work Papers Comments

 

Validation

Validation of the ensemble is done by comparing the ensemble track forecasts with the best track, other forecast aids, climatology and persistance (CLIPER), etc. Validation of the ensemble-computed strike probabilities should ultimately involve the use of the Brier Skill Score and the Relative Operating Characteristics (ROC).

  • Seasonal verification of total ensemble mean and subensemble mean track forecasts for 2001, 2002, and 2003 seasons:
    • Comparison with control member
    • Comparison with randomly selected subensembles of various sizes
    • Comparison with other barotropic models
    • Comparison with the GFS ensemble track forecasts
    • Comparison with full physics and global models
    • Comparison with other ensembles (GUNA, AEMN)

     

  • Verification of explicitly computed strike probabilities
    • Comparison with NHC strike probabilities (not done yet)
    • Comparison with strike probabilities estimated from the GFS ensembles
    • Comparison with strike probabilities estimated from hybrid ensembles

 

  • Determination of the ensemble spread vs. ensemble error relationship

 

  • Evaluation of the reliability of the ensemble envelope

   

 

On to the Future Work section . . .